基于车辆轨迹数据的交叉口危险驾驶行为预测

发布时间：2018-06-14 07:13

  本文选题：危险行为预测 + 车辆轨迹预测　； 参考：《同济大学学报(自然科学版)》2017年10期

【摘要】：基于大量实证车辆轨迹数据建立了信号控制交叉口相位切换期间轨迹预测模型和驾驶员心理决策过程预测模型,并在两个模型的基础上建立了驾驶员危险行为识别的规则,三者共同构成了驾驶员危险行为预测方法.经检验,预测方法的整体精度达到88.89%.该预测方法将驾驶员临近交叉口时的隐藏心理决策过程作为一个重要变量运用到危险行为预测方法中,并实现了对车辆个体动态化的轨迹预测.
[Abstract]:Based on a large number of empirical vehicle trajectory data, the prediction model of trajectory during phase switching of signal-controlled intersection and the prediction model of driver's psychological decision-making process are established, and the identification rules of driver's dangerous behavior are established on the basis of the two models. The three methods together constitute the prediction method of driver's dangerous behavior. The overall accuracy of the prediction method is 88.89. The hidden psychological decision-making process of the driver near the intersection is used as an important variable in the prediction method of dangerous behavior, and the trajectory prediction of individual vehicle is realized.
【作者单位】： 同济大学道路与交通工程教育部重点实验室;
【基金】：国家自然科学基金(61673302)
【分类号】：U491.25
，

本文编号：2016606